Too little damping lets a knob turn by itself under vibration

What it is

In a car, on a boat, in a pocket, the chassis shakes and the knob walks a step at a time: volume climbs, the channel jumps. Vibration walk from too little damping is stability after release: detent events are still there, but they do not stop the environment from pushing the shaft over a notch. It is not “too few steps,” and not whether steps are pleasant to count. It is whether the value changes with no one operating it.

Why it happens

Vibration puts a random torque on the shaft. Damping (grease, hysteresis, an O-ring) eats angular velocity; notch depth of a pawl sets a torque that must be exceeded. When both sit below peaks in the vibration spectrum, the shaft walks a step at a time, or accumulates angle on a smooth stretch. An endless encoder has no stop, so whatever it walks, software faithfully records. An end-stopped pot only stops at the end; values in between still drift. Low-frequency, large-amplitude driving, broadband pocket shake, and music vibration on a speaker panel are different spectra, and different knobs will walk. The human criterion is “I did not touch it, it must not move”—harder than “is it light to turn.”

Studying it

Put the assembled device on the target vibration spectrum (automotive, portable drop, speaker panel), knob in each mounting attitude, and measure angular displacement with no one operating it.

Independent variables: damping torque, notch depth, shaft inertia, mounting attitude, spectrum, presence of a software minimum-velocity gate. Dependent measures: steps walked per unit time, whether walk is one-way, whether software logs vibration as input.

A desktop feel evaluation of twisting by hand will not catch walk. A software gate can block slow crawl and also filter a slow intentional trim; report false blocks separately. Temperature changes grease viscosity; run hot and cold.

Where it stops holding

A fixed desktop with no vibration source: too little damping is only a slippery feel, not necessarily walk. Scenes that must turn with one light finger (some disabilities, very thick gloves) fight “block automotive vibration”; you may need a different mount or a lock. Magnetic hysteresis damping is strong at rest and relatively slippery in motion, unlike viscous grease’s speed curve; do not mash them into one “damping amount.” Software that classifies vibration as “not a person” can help, at the cost of delay and killed trims. Safety-related values (minimum brightness, a radio transmit) changed by walk must be stopped by a hardware threshold, not only by after-the-fact detection.

Applying it

• Set minimum holding torque from the vibration spectrum the product will see, so unattended walk is zero—not from whether it feels light on a desk.
• A software gate on an endless encoder is a second line; the first is still mechanical damping or notch depth.
• Tighten the spec when the shaft sits along the main vibration axis; do not assume a lab’s level desk.
• Verify: run each attitude on the target spectrum for the specified duration; the knob must not walk. Then, at the same damping, do a slow intentional trim and confirm the gate did not eat it. Retest hot and cold. If walk would change a safety-related value, add a mechanical lock or a deeper notch, not only a log.

Related

• Same group: C10.10.1 Detents per revolution cap the granularity a single turn can achieve · C10.10.2 Mechanical detents come from a physical pawl; electronic detents are motor-simulated, and they feel different · C10.10.4 The resistance curve between detents decides whether people can count steps by feel alone
• Adjacent: C10.02 Knobs and Continuous Adjustment · C10.16 Mechanical Reliability and Lifetime
• Search: vibration walk · knob damping · holding torque